Richard Gomulkiewicz

 

As I understand it I have the honor, of which I am most definetly not worthy, of being the first Gomulkiewicz Keynote Speaker. Many of you here knew him but for those prospective students who didn’t that good fortune to meet him, he was a PI here at WSU but sadly passed away in 2023. His was an theoretical population geneticist and had a reputation for being tough. More than one of my fellow grad students warned me away from chosing him to be on my commitee during me first year. I am so grateful that I chose not to listen to that advice.

Mathematical Geneology

 

Outline

  1. About me
  2. Adaptation in stochastic environments
  3. Some of my ML projects
    • Regression-based Machine Learning (WSU)
    • Classification-based Machine Learning (UIdaho)
    • Generative AI-based Machine Learning (EHA)
  4. Closing
  5. Questions

About Me

Adaptation in Stochastic Environments

  • Bet-Hedging
    • Definition: Increasing variance in fitness at the expense of lower average fitness to ensure survival across unpredictable conditions.
    • Examples: Plants producing seeds with staggered germination times.
  • Life History Adjustments
    • r-selection: High fecundity with low parental investment (e.g., many offspring).
    • K-selection: Fewer offspring with higher investment (e.g., longevity, parental care).
    • Dormancy: Delayed development (e.g., seed banks).
  • Boosting Genetic Diversity
    • Role: Buffers populations against fluctuations, ensuring some individuals thrive in any condition.
    • Mechanisms: Variation of mutation rate, sexual reproduction, dispersal / gene flow.
  • Phenotypic Plasticity
    • Definition: Non-parallel reaction norms among individuals with different genotypes in response to different environmental conditions. GxE interaction.
    • Examples: Seasonal coat color changes in animals; metabolic flexibility in plants.

These Hypotheses can Be Tested

“The fitness of a lineage in a fluctuating environment is the time average of its fitness over the sequence of static conditions it encounters.”1

Different Kinds of Mean

 

Machine Learning

XKCD’s definition

 

Large Language Models

Modeling Language

 

https://medium.com/data-science-at-microsoft/how-large-language-models-work-91c362f5b78f

Regression-based Machine Learning

Can we reconstruct historical of range expansion routes?

 

Time Distance Of Arrival (TDOA)

 

Westward range expansion from middle latitudes explains the Mississippi River discontinuity in a forest herb of eastern North America2

Identifying the refugia

Sweeps in space: leveraging geographic data to identify beneficial alleles in Anopheles gambiae3

Classification-based Machine Learning

Land Cover Classification

Predicting the fine-scale spatial distribution of zoonotic reservoirs using computer vision4

Zoonotic Reservoirs

Predicting the fine-scale spatial distribution of zoonotic reservoirs using computer vision4

Fine-scale Predictors

Predicting the fine-scale spatial distribution of zoonotic reservoirs using computer vision4

Generative AI-based Machine Learning

EcoHealth Alliance is a nonprofit dedicated to using science-based solutions to prevent pandemics and promote conservation

World Organisation for Animal Health (WOAH)

Quantifying Underreporting in Wildlife Disease Surveillance

Automating a Lit Search

Quantifying Underreporting in Wildlife Disease Surveillance

Correcting for LLM errors using simulation extrapolation (SIMEX)

Do I think science is a viable career path?

Yes

But sometimes if feels like

Adaptation in a Stochastic Research Environment

Biologists are plastic!

For example

Almost all of my academic friends, have all worked in both academic and non-academic environments at some point

I know graduates who are currently:

  • Program manager
  • Database management
  • Data librarian
  • Data scientist
  • Mathematical Statistician
  • Researcher at the Smithsonian
  • IT / AWS specialist
  • Public health educator
  • Consultant
  • Environmental health professional
  • Biotechnology startup researcher

Adaptation in Stochastic Research Research Environments

  • Bet-Hedging
    • Definition: Increasing variance in fitness at the expense of lower average fitness to ensure survival across unpredictable conditions.
    • Examples: Plants producing seeds with staggered germination times.
  • Life History Adjustments
    • r-selection: High fecundity with low parental investment (e.g., many offspring).
    • K-selection: Fewer offspring with higher investment (e.g., longevity, parental care).
    • Dormancy: Delayed development (e.g., seed banks).
  • Boosting Genetic Diversity
    • Role: Buffers populations against fluctuations, ensuring some individuals thrive in any condition.
    • Mechanisms: Variation of mutation rate, sexual reproduction, dispersal / gene flow.
  • Phenotypic Plasticity
    • Definition: Non-parallel reaction norms among individuals with different genotypes in response to different environmental conditions. GxE interaction.
    • Examples: Seasonal coat color changes in animals; metabolic flexibility in plants.

But what about being replaced by AI?

But what about being replaced by AI?

But what about being replaced by AI?

More importantly, an generative AI is a based on massively complicated statistical engines designed to answer the question, “what should I say next?”

But what about being replaced by AI?

 

    If you make a complicated model of nature,

But what about being replaced by AI?

 

    If you make a complicated model of nature,
    now you have two things you don’t understand.
          -Alan Hastings

But what about being replaced by AI?

But what about being replaced by AI?

 

    AI is a tool to help people answer questions. It’s as useful as the p-value. But we are the people who ask questions.

My advice

Doing science beyond grad school involves

  1. Leveraging your network
  2. Managing resources
  3. Looking to the future

Why aren’t you doing that now?

References

1.
Abreu, C. I., Mathur, S. & Petrov, D. A. Environmental memory alters the fitness effects of adaptive mutations in fluctuating environments. Nat. Ecol. Evol. 8, 1760–1775 (2024).
2.
Prior, C. J., Layman, N. C., Koski, M. H., Galloway, L. F. & Busch, J. W. Westward range expansion from middle latitudes explains the mississippi river discontinuity in a forest herb of eastern north america. Molecular Ecology 29, 4473–4486 (2020).
3.
Rehmann, C. T., Small, S. T., Ralph, P. L. & Kern, A. D. Sweeps in space: Leveraging geographic data to identify beneficial alleles in anopheles gambiae. bioRxiv (2025) doi:10.1101/2025.02.07.637123.
4.
Layman, N. C. et al. Predicting the fine‐scale spatial distribution of zoonotic reservoirs using computer vision. Ecology Letters 26, 1974–1986 (2023).